diff mbox series

[03/17] RDMA/core: Introduce IB_MR_TYPE_PI and ib_alloc_mr_integrity API

Message ID 1549560811-8655-4-git-send-email-maxg@mellanox.com (mailing list archive)
State Superseded
Headers show
Series Introduce new API for T10-PI offload | expand

Commit Message

Max Gurtovoy Feb. 7, 2019, 5:33 p.m. UTC
From: Israel Rukshin <israelr@mellanox.com>

This is a preparation for signature verbs API re-design. In the new
design a single MR with IB_MR_TYPE_PI type will be used to perform
the needed mapping for PI (protection information) operations.

Signed-off-by: Israel Rukshin <israelr@mellanox.com>
Signed-off-by: Max Gurtovoy <maxg@mellanox.com>
---
 drivers/infiniband/core/device.c |  1 +
 drivers/infiniband/core/verbs.c  | 46 ++++++++++++++++++++++++++++++++++++++++
 include/rdma/ib_verbs.h          | 11 ++++++++++
 3 files changed, 58 insertions(+)

Comments

Sagi Grimberg Feb. 7, 2019, 6:01 p.m. UTC | #1
> From: Israel Rukshin <israelr@mellanox.com>
> 
> This is a preparation for signature verbs API re-design. In the new
> design a single MR with IB_MR_TYPE_PI type will be used to perform
> the needed mapping for PI (protection information) operations.

One nit, lets keep mr type and allocation api consistent:
IB_MR_TYPE_INTEGRITY
ib_alloc_mr_integrity

Other than that,
Reviewed-by: Sagi Grimberg <sagi@grimberg.me>
Christoph Hellwig Feb. 12, 2019, 3:59 p.m. UTC | #2
> @@ -1982,6 +1982,9 @@ struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
>  	if (!pd->device->ops.alloc_mr)
>  		return ERR_PTR(-EOPNOTSUPP);
>  
> +	if (WARN_ON_ONCE(mr_type == IB_MR_TYPE_PI))
> +		return ERR_PTR(-EINVAL);
> +

So why is IB_MR_TYPE_PI a separate function, but IB_MR_TYPE_SG_GAPS
is not?

I think we either want one alloc/free helper per type, or we have
to extend ib_alloc_mr with an arguments structure or something similar.
Max Gurtovoy Feb. 13, 2019, 12:41 a.m. UTC | #3
On 2/12/2019 5:59 PM, Christoph Hellwig wrote:
>> @@ -1982,6 +1982,9 @@ struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
>>   	if (!pd->device->ops.alloc_mr)
>>   		return ERR_PTR(-EOPNOTSUPP);
>>   
>> +	if (WARN_ON_ONCE(mr_type == IB_MR_TYPE_PI))
>> +		return ERR_PTR(-EINVAL);
>> +
> So why is IB_MR_TYPE_PI a separate function, but IB_MR_TYPE_SG_GAPS
> is not?
>
> I think we either want one alloc/free helper per type, or we have
> to extend ib_alloc_mr with an arguments structure or something similar.

my original commit modified the ib_alloc_mr to:

struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
                           enum ib_mr_type mr_type,
-                         u32 max_num_sg)
+                         u32 max_num_sg, u32 max_num_meta_sg)


and I also prefer to have 1 allocation function but I guess we should 
use the style that the RDMA maintainers prefer and this is the review I got.

So Jason/Leon/Doug in case you prefer to extend ib_alloc_mr as I 
originally did, I can do it for V3.
Jason Gunthorpe Feb. 14, 2019, 5:53 p.m. UTC | #4
On Wed, Feb 13, 2019 at 02:41:49AM +0200, Max Gurtovoy wrote:
> 
> On 2/12/2019 5:59 PM, Christoph Hellwig wrote:
> > > @@ -1982,6 +1982,9 @@ struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
> > >   	if (!pd->device->ops.alloc_mr)
> > >   		return ERR_PTR(-EOPNOTSUPP);
> > > +	if (WARN_ON_ONCE(mr_type == IB_MR_TYPE_PI))
> > > +		return ERR_PTR(-EINVAL);
> > > +
> > So why is IB_MR_TYPE_PI a separate function, but IB_MR_TYPE_SG_GAPS
> > is not?
> > 
> > I think we either want one alloc/free helper per type, or we have
> > to extend ib_alloc_mr with an arguments structure or something similar.
> 
> my original commit modified the ib_alloc_mr to:
> 
> struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
>                           enum ib_mr_type mr_type,
> -                         u32 max_num_sg)
> +                         u32 max_num_sg, u32 max_num_meta_sg)
> 
> 
> and I also prefer to have 1 allocation function but I guess we should use
> the style that the RDMA maintainers prefer and this is the review I got.
> 
> So Jason/Leon/Doug in case you prefer to extend ib_alloc_mr as I originally
> did, I can do it for V3.

These monster 'do everything' multiplexor functions have turned out to
be a maintenance pain in the past. I'm not sure MR will turn out any
better..

So, CH's 'one alloc/free helper per type' seems like a better
direction. At least the API should be understandable in that
configuration.

Jason
diff mbox series

Patch

diff --git a/drivers/infiniband/core/device.c b/drivers/infiniband/core/device.c
index 238ec42778ef..bc4c7c5c305b 100644
--- a/drivers/infiniband/core/device.c
+++ b/drivers/infiniband/core/device.c
@@ -1244,6 +1244,7 @@  void ib_set_device_ops(struct ib_device *dev, const struct ib_device_ops *ops)
 	SET_DEVICE_OP(dev_ops, alloc_fmr);
 	SET_DEVICE_OP(dev_ops, alloc_hw_stats);
 	SET_DEVICE_OP(dev_ops, alloc_mr);
+	SET_DEVICE_OP(dev_ops, alloc_mr_integrity);
 	SET_DEVICE_OP(dev_ops, alloc_mw);
 	SET_DEVICE_OP(dev_ops, alloc_pd);
 	SET_DEVICE_OP(dev_ops, alloc_rdma_netdev);
diff --git a/drivers/infiniband/core/verbs.c b/drivers/infiniband/core/verbs.c
index 4588b933d4b4..047a51bd3e79 100644
--- a/drivers/infiniband/core/verbs.c
+++ b/drivers/infiniband/core/verbs.c
@@ -1982,6 +1982,9 @@  struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
 	if (!pd->device->ops.alloc_mr)
 		return ERR_PTR(-EOPNOTSUPP);
 
+	if (WARN_ON_ONCE(mr_type == IB_MR_TYPE_PI))
+		return ERR_PTR(-EINVAL);
+
 	mr = pd->device->ops.alloc_mr(pd, mr_type, max_num_sg);
 	if (!IS_ERR(mr)) {
 		mr->device  = pd->device;
@@ -1999,6 +2002,49 @@  struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
 }
 EXPORT_SYMBOL(ib_alloc_mr);
 
+/**
+ * ib_alloc_mr_integrity() - Allocates an integrity memory region
+ * @pd:                      protection domain associated with the region
+ * @max_num_data_sg:         maximum data sg entries available for registration
+ * @max_num_meta_sg:         maximum metadata sg entries available for
+ *                           registration
+ *
+ * Notes:
+ * Memory registration page/sg lists must not exceed max_num_sg,
+ * also the integrity page/sg lists must not exceed max_num_meta_sg.
+ *
+ */
+struct ib_mr *ib_alloc_mr_integrity(struct ib_pd *pd,
+				    u32 max_num_data_sg,
+				    u32 max_num_meta_sg)
+{
+	struct ib_mr *mr;
+
+	if (!pd->device->ops.alloc_mr_integrity)
+		return ERR_PTR(-EOPNOTSUPP);
+
+	if (!max_num_meta_sg)
+		return ERR_PTR(-EINVAL);
+
+	mr = pd->device->ops.alloc_mr_integrity(pd, max_num_data_sg,
+						max_num_meta_sg);
+	if (IS_ERR(mr))
+		return mr;
+
+	mr->device = pd->device;
+	mr->pd = pd;
+	mr->dm = NULL;
+	mr->uobject = NULL;
+	atomic_inc(&pd->usecnt);
+	mr->need_inval = false;
+	mr->res.type = RDMA_RESTRACK_MR;
+	rdma_restrack_kadd(&mr->res);
+	mr->type = IB_MR_TYPE_PI;
+
+	return mr;
+}
+EXPORT_SYMBOL(ib_alloc_mr_integrity);
+
 /* "Fast" memory regions */
 
 struct ib_fmr *ib_alloc_fmr(struct ib_pd *pd,
diff --git a/include/rdma/ib_verbs.h b/include/rdma/ib_verbs.h
index 42a9e297c21f..8edba47f272e 100644
--- a/include/rdma/ib_verbs.h
+++ b/include/rdma/ib_verbs.h
@@ -756,11 +756,15 @@  __attribute_const__ int ib_rate_to_mbps(enum ib_rate rate);
  *                            register any arbitrary sg lists (without
  *                            the normal mr constraints - see
  *                            ib_map_mr_sg)
+ * @IB_MR_TYPE_PI:            memory region that is used for
+ *                            PI (protection information) operations
+
  */
 enum ib_mr_type {
 	IB_MR_TYPE_MEM_REG,
 	IB_MR_TYPE_SIGNATURE,
 	IB_MR_TYPE_SG_GAPS,
+	IB_MR_TYPE_PI,
 };
 
 enum ib_mr_status_check {
@@ -2306,6 +2310,9 @@  struct ib_device_ops {
 	int (*dereg_mr)(struct ib_mr *mr);
 	struct ib_mr *(*alloc_mr)(struct ib_pd *pd, enum ib_mr_type mr_type,
 				  u32 max_num_sg);
+	struct ib_mr *(*alloc_mr_integrity)(struct ib_pd *pd,
+					    u32 max_num_data_sg,
+					    u32 max_num_meta_sg);
 	int (*advise_mr)(struct ib_pd *pd,
 			 enum ib_uverbs_advise_mr_advice advice, u32 flags,
 			 struct ib_sge *sg_list, u32 num_sge,
@@ -3677,6 +3684,10 @@  struct ib_mr *ib_alloc_mr(struct ib_pd *pd,
 			  enum ib_mr_type mr_type,
 			  u32 max_num_sg);
 
+struct ib_mr *ib_alloc_mr_integrity(struct ib_pd *pd,
+				    u32 max_num_data_sg,
+				    u32 max_num_meta_sg);
+
 /**
  * ib_update_fast_reg_key - updates the key portion of the fast_reg MR
  *   R_Key and L_Key.